Design of Intelligent Controllers Using Online-Trained Fuzzy Neural Networks for the UPFC

Abstract

With the trend toward deregulating the power industry, it has been expected that future power systems will need to provide sufficient, secure, high-quality and real-time controllable electric power to various load centers. It is envisaged that flexible AC transmission systems (FACTS) devices [1,2] or similar power flow controllers are going to play a critical role in operating the new type of power systems under such a complex operating environment. Basically, using state-of-the-art communication technologies, high-power converters and properly designed controllers, FACTS devices can offer great control flexibility in power system operations. It has been proved that a number of important control functions and optimization issues, which were unattainable in the past, can now be easily realized by utilizing these high-performance power flow controllers with some properly developed control algorithms. One of the crucial objectives concerning FACTS device applications in modern power systems is real-time load flow control. Of the reported FACTS devices, it has been well accepted that the unified power flow controller (UPFC) is the most versatile and powerful one [3, 4] that can provide effective means for controlling the power flow and improving the transient stability of a power network [5,6]. However, the UPFC has multiple operating modes and complex system dynamics that require advanced controllers to achieve satisfactory performances. In the literature, a number of conventional controllers have been proposed for the UPFC to perform various power system control functions; however, most of them are linear controllers designed on a linearized model of the controlled power system. To achieve robust control effects and better dynamic control performance some advanced controllers with adaptive features and the ability to learn will be required.